1 |
dpavlin |
2 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
package MWS; |
4 |
|
|
|
5 |
|
|
use strict; |
6 |
|
|
use warnings; |
7 |
dpavlin |
41 |
|
8 |
|
|
use lib '.'; |
9 |
|
|
|
10 |
dpavlin |
2 |
use Carp; |
11 |
dpavlin |
3 |
use Mail::Box::Manager; |
12 |
dpavlin |
4 |
use Config::IniFiles; |
13 |
dpavlin |
13 |
use POSIX qw(strftime); |
14 |
dpavlin |
14 |
use Text::Autoformat; |
15 |
dpavlin |
16 |
use Text::Iconv; |
16 |
dpavlin |
19 |
use Text::Unaccent; |
17 |
dpavlin |
20 |
use Date::Parse; |
18 |
|
|
use POSIX qw(strftime); |
19 |
dpavlin |
21 |
use MIME::Base64; |
20 |
dpavlin |
14 |
|
21 |
dpavlin |
2 |
our $VERSION = '1.00'; |
22 |
|
|
|
23 |
dpavlin |
30 |
my $debug = 1; |
24 |
dpavlin |
2 |
|
25 |
dpavlin |
42 |
my @counters = qw(from to cc bcc folder); |
26 |
|
|
|
27 |
dpavlin |
2 |
sub new { |
28 |
|
|
my $class = shift; |
29 |
dpavlin |
41 |
my $self = {@_}; |
30 |
dpavlin |
2 |
bless($self, $class); |
31 |
|
|
|
32 |
|
|
|
33 |
dpavlin |
41 |
my $config_name = $self->{config_file} || croak "need config_file"; |
34 |
|
|
$config_name =~ s/\.conf.*$//; |
35 |
|
|
$self->{config_name} = $config_name; |
36 |
dpavlin |
2 |
|
37 |
dpavlin |
41 |
$self->{config} = new Config::IniFiles( -file => $self->{config_file} ); |
38 |
dpavlin |
4 |
|
39 |
dpavlin |
27 |
my $index_dir = $self->{config}->val('global', 'index') || croak "can't find [index] section in config file with path of index"; |
40 |
|
|
|
41 |
dpavlin |
12 |
$self->{mgr} = Mail::Box::Manager->new(access => 'r'); |
42 |
dpavlin |
27 |
$self->{index_dir} = $index_dir; |
43 |
dpavlin |
4 |
|
44 |
dpavlin |
2 |
# placeholder for opened folders |
45 |
|
|
$self->{folder} = {}; |
46 |
|
|
|
47 |
dpavlin |
14 |
$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin'); |
48 |
dpavlin |
17 |
$self->{max_results} = $self->{config}->val('global', 'max_results') || 100; |
49 |
dpavlin |
19 |
$self->reset_counters; |
50 |
dpavlin |
14 |
|
51 |
dpavlin |
2 |
return $self; |
52 |
|
|
} |
53 |
|
|
|
54 |
dpavlin |
19 |
sub normalize_string { |
55 |
|
|
my $self = shift; |
56 |
|
|
|
57 |
|
|
my $v = shift || return; |
58 |
|
|
|
59 |
|
|
$v = unac_string('ISO-8859-2', $v); |
60 |
|
|
$v = join('',sort split(/\s+/,$v)); |
61 |
|
|
$v =~ s/\W+//g; |
62 |
|
|
|
63 |
dpavlin |
23 |
return lc($v); |
64 |
dpavlin |
19 |
} |
65 |
|
|
|
66 |
|
|
# reset tables for search results |
67 |
|
|
sub reset_counters { |
68 |
|
|
my $self = shift; |
69 |
|
|
|
70 |
|
|
$self->{counter} = {}; |
71 |
|
|
|
72 |
|
|
# foreach my $c (qw(thread from to cc bcc lists links att)) { |
73 |
|
|
# $self->{counter}->{$c} = {}; |
74 |
|
|
# } |
75 |
|
|
|
76 |
|
|
} |
77 |
|
|
|
78 |
|
|
sub add_counter($$) { |
79 |
|
|
my $self = shift; |
80 |
|
|
|
81 |
|
|
my ($c,$v) = @_; |
82 |
|
|
my $k = $self->normalize_string($v); |
83 |
|
|
|
84 |
dpavlin |
42 |
$self->{counter}->{$c}->{$k}->{name} = $v || return; |
85 |
dpavlin |
19 |
return $self->{counter}->{$c}->{$k}->{usage}++; |
86 |
|
|
} |
87 |
|
|
|
88 |
dpavlin |
20 |
sub yyyymmdd { |
89 |
|
|
my $self = shift; |
90 |
|
|
|
91 |
|
|
my $t = shift || time; |
92 |
|
|
|
93 |
|
|
my (undef,undef,undef,$dd,$mm,$yyyy) = localtime($t); |
94 |
|
|
$mm++; |
95 |
|
|
$yyyy+=1900; |
96 |
|
|
return ($yyyy,$mm,$dd); |
97 |
|
|
} |
98 |
|
|
|
99 |
|
|
sub fmtdate { |
100 |
|
|
my $self = shift; |
101 |
|
|
|
102 |
|
|
my @out; |
103 |
|
|
my @formats = qw(%04d %02d %02d); |
104 |
|
|
while (my $v = shift) { |
105 |
|
|
my $f = shift @formats; |
106 |
|
|
push @out, sprintf($f, $v); |
107 |
|
|
} |
108 |
|
|
|
109 |
dpavlin |
30 |
print STDERR "fmtdate: ",join('|',@out),"\n" if ($debug == 2); |
110 |
dpavlin |
20 |
|
111 |
|
|
return (wantarray ? @out : join("-",@out)); |
112 |
|
|
} |
113 |
|
|
|
114 |
|
|
sub add_counter_calendar($) { |
115 |
|
|
my $self = shift; |
116 |
|
|
|
117 |
|
|
my $t = shift || croak "add_counter_calendar without argument!"; |
118 |
|
|
|
119 |
|
|
my ($yyyy,$mm,$dd) = $self->fmtdate($self->yyyymmdd($t)); |
120 |
|
|
|
121 |
|
|
return $self->{counter}->{calendar}->{"$yyyy-$mm"}->{$dd}++; |
122 |
|
|
} |
123 |
|
|
|
124 |
|
|
|
125 |
dpavlin |
19 |
sub counter { |
126 |
|
|
my $self = shift; |
127 |
|
|
|
128 |
|
|
my $c = shift || return; |
129 |
|
|
|
130 |
|
|
return if (! $self->{counter}->{$c}); |
131 |
|
|
|
132 |
|
|
return $self->{counter}->{$c}; |
133 |
|
|
} |
134 |
|
|
|
135 |
dpavlin |
14 |
sub mbox_name2path { |
136 |
|
|
my $self = shift; |
137 |
|
|
|
138 |
|
|
my $mbox = shift || croak "folder_name2path needs mbox name"; |
139 |
|
|
|
140 |
|
|
return $self->{config}->val('folders', $mbox) || croak "comeone removed folder $mbox from config?"; |
141 |
|
|
} |
142 |
|
|
|
143 |
dpavlin |
7 |
sub open_folder { |
144 |
dpavlin |
2 |
my $self = shift; |
145 |
|
|
|
146 |
dpavlin |
7 |
my $mbox = shift || croak "open_folder needs mbox name"; |
147 |
dpavlin |
2 |
|
148 |
|
|
if (! $self->{folder}->{$mbox}) { |
149 |
dpavlin |
14 |
my $mbox_path = $self->mbox_name2path($mbox); |
150 |
|
|
|
151 |
dpavlin |
13 |
print STDERR "about to open_folder($mbox)\n" if ($debug == 2); |
152 |
dpavlin |
14 |
|
153 |
dpavlin |
4 |
$self->{folder}->{$mbox} = $self->{mgr}->open($mbox_path) || croak "can't open folder $mbox at '$mbox_path': $!"; |
154 |
dpavlin |
14 |
|
155 |
|
|
print STDERR "open_folder($mbox) ok\n" if ($debug); |
156 |
dpavlin |
2 |
} |
157 |
|
|
|
158 |
dpavlin |
16 |
$self->{fetch_count} = 0; |
159 |
|
|
|
160 |
dpavlin |
7 |
return $self->{folder}->{$mbox}; |
161 |
|
|
|
162 |
|
|
} |
163 |
|
|
|
164 |
dpavlin |
14 |
sub close_folder { |
165 |
|
|
my $self = shift; |
166 |
|
|
|
167 |
|
|
my $mbox = shift || croak "open_folder needs mbox name"; |
168 |
|
|
|
169 |
dpavlin |
16 |
$self->{folder}->{$mbox}->close(write => 'NEVER') || croak "can't close folder $mbox"; |
170 |
|
|
|
171 |
|
|
# XXX this is rather agressive!!! |
172 |
|
|
$self->{folder} = {}; |
173 |
|
|
return |
174 |
dpavlin |
14 |
} |
175 |
|
|
|
176 |
dpavlin |
7 |
sub fetch_message { |
177 |
|
|
my $self = shift; |
178 |
|
|
|
179 |
|
|
my $mbox_id = shift || die "need mbox_id!"; |
180 |
|
|
my ($mbox,$id) = split(/ /,$mbox_id); |
181 |
|
|
|
182 |
|
|
# return message with ID |
183 |
dpavlin |
13 |
print STDERR "fetch $id from $mbox\n" if ($debug); |
184 |
dpavlin |
16 |
|
185 |
|
|
if ($self->{fetch_count}++ > 100) { |
186 |
|
|
$self->close_folder($mbox); |
187 |
|
|
print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n"; |
188 |
|
|
} |
189 |
|
|
|
190 |
dpavlin |
17 |
my $msg = $self->open_folder($mbox)->find($id); |
191 |
|
|
if ($msg) { |
192 |
|
|
return $msg; |
193 |
|
|
} else { |
194 |
dpavlin |
2 |
print STDERR "can't find message $id in $mbox. Time to re-index?\n"; |
195 |
dpavlin |
17 |
return; |
196 |
|
|
} |
197 |
dpavlin |
2 |
} |
198 |
|
|
|
199 |
|
|
|
200 |
|
|
sub search { |
201 |
|
|
my $self = shift; |
202 |
|
|
|
203 |
dpavlin |
21 |
carp "search called without argument!" if (! @_); |
204 |
dpavlin |
2 |
|
205 |
dpavlin |
19 |
$self->reset_counters; |
206 |
|
|
|
207 |
dpavlin |
21 |
print STDERR "search(",join(" ",@_),")\n" if ($debug == 2); |
208 |
|
|
my @index_ids = $self->search_index(@_); |
209 |
dpavlin |
2 |
|
210 |
|
|
$self->{'index_ids'} = \@index_ids; |
211 |
|
|
|
212 |
dpavlin |
17 |
#my $results = $#index_ids + 1; |
213 |
|
|
#$self->{'results'} = $results; |
214 |
|
|
|
215 |
|
|
my $results = $self->{'total_hits'} || ($#index_ids + 1); |
216 |
dpavlin |
2 |
|
217 |
|
|
$self->{'curr_result'} = 0; |
218 |
|
|
|
219 |
dpavlin |
23 |
$self->reset_counters; |
220 |
|
|
|
221 |
dpavlin |
13 |
print STDERR "$results results\n" if ($debug == 2); |
222 |
|
|
|
223 |
dpavlin |
2 |
return $results || 'error'; |
224 |
|
|
} |
225 |
|
|
|
226 |
dpavlin |
16 |
sub decode_qp($) { |
227 |
|
|
my $self = shift; |
228 |
|
|
|
229 |
|
|
my $tmp = shift || return; |
230 |
|
|
|
231 |
dpavlin |
21 |
sub decode($$$) { |
232 |
|
|
my ($cp,$enc,$qp) = @_; |
233 |
|
|
|
234 |
|
|
print STDERR "decode($cp,$qp) -> " if ($debug == 2); |
235 |
|
|
|
236 |
|
|
if (uc($enc) eq "Q") { |
237 |
|
|
$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg; |
238 |
|
|
$qp =~ s/_/ /g; |
239 |
|
|
} elsif (uc($enc) eq "B") { |
240 |
|
|
$qp = decode_base64($qp); |
241 |
|
|
} else { |
242 |
|
|
croak "unsupported encoding '$enc' in decode_qp\n"; |
243 |
|
|
return $qp; |
244 |
|
|
} |
245 |
|
|
|
246 |
dpavlin |
22 |
print STDERR "$qp\n" if ($debug == 2); |
247 |
dpavlin |
21 |
|
248 |
dpavlin |
20 |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
249 |
|
|
return $iconv->convert($qp) || ''; |
250 |
dpavlin |
16 |
} |
251 |
|
|
|
252 |
dpavlin |
21 |
$tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige; |
253 |
dpavlin |
17 |
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
254 |
dpavlin |
20 |
#print STDERR "$tmp\n" if ($debug == 2); |
255 |
dpavlin |
16 |
return $tmp; |
256 |
|
|
} |
257 |
|
|
|
258 |
dpavlin |
2 |
sub unroll($$$) { |
259 |
dpavlin |
13 |
my $self = shift; |
260 |
|
|
|
261 |
dpavlin |
2 |
my ($message,$part,$sub) = @_; |
262 |
|
|
|
263 |
|
|
my @arr; |
264 |
|
|
|
265 |
dpavlin |
17 |
return if (! $message->$part); |
266 |
|
|
|
267 |
dpavlin |
2 |
foreach my $from ($message->$part) { |
268 |
dpavlin |
16 |
my $tmp = $from->$sub || next; |
269 |
|
|
|
270 |
|
|
$tmp = $self->decode_qp($tmp); |
271 |
|
|
push @arr, $tmp; |
272 |
dpavlin |
2 |
} |
273 |
dpavlin |
14 |
|
274 |
|
|
return @arr; |
275 |
dpavlin |
2 |
} |
276 |
dpavlin |
4 |
|
277 |
|
|
sub fetch_all_results { |
278 |
|
|
my $self = shift; |
279 |
|
|
|
280 |
|
|
croak "results called before search!" if (! $self->{'index_ids'}); |
281 |
|
|
|
282 |
dpavlin |
14 |
print STDERR "fetch_all_results [",scalar @{$self->{'index_ids'}},"]\n" if ($debug == 2); |
283 |
dpavlin |
13 |
|
284 |
dpavlin |
4 |
my @arr; |
285 |
|
|
|
286 |
|
|
foreach my $id (@{$self->{'index_ids'}}) { |
287 |
|
|
push @arr, $self->fetch_result_by_id($id); |
288 |
|
|
} |
289 |
|
|
|
290 |
dpavlin |
19 |
|
291 |
dpavlin |
4 |
return @arr; |
292 |
|
|
} |
293 |
|
|
|
294 |
dpavlin |
2 |
sub fetch_result { |
295 |
|
|
my $self = shift; |
296 |
|
|
|
297 |
dpavlin |
3 |
my $args = {@_}; |
298 |
|
|
|
299 |
dpavlin |
2 |
croak "results called before search!" if (! $self->{'index_ids'}); |
300 |
|
|
|
301 |
|
|
my $curr = $self->{'curr_result'}++; |
302 |
|
|
|
303 |
|
|
my $id = $self->{'index_ids'}->[$curr]; |
304 |
dpavlin |
14 |
|
305 |
|
|
print STDERR "fetch_result: $curr = $id\n" if ($debug == 2); |
306 |
dpavlin |
2 |
|
307 |
dpavlin |
4 |
return $self->fetch_result_by_id($id); |
308 |
|
|
} |
309 |
dpavlin |
2 |
|
310 |
dpavlin |
7 |
sub plain_text_body { |
311 |
|
|
my $self = shift; |
312 |
|
|
my $message = shift || croak "plain_text_body needs message!"; |
313 |
|
|
|
314 |
dpavlin |
14 |
my $body; |
315 |
|
|
|
316 |
dpavlin |
7 |
if (! $message->isMultipart) { |
317 |
dpavlin |
14 |
$body = $message->decoded->string; |
318 |
dpavlin |
7 |
} else { |
319 |
|
|
foreach my $part ($message->parts) { |
320 |
|
|
if ($part->body->mimeType eq 'text/plain') { |
321 |
dpavlin |
14 |
$body = $part->decoded->string; |
322 |
|
|
last; |
323 |
dpavlin |
7 |
} |
324 |
|
|
} |
325 |
|
|
} |
326 |
dpavlin |
14 |
|
327 |
dpavlin |
25 |
if (! $body) { |
328 |
dpavlin |
26 |
$body = "[plain/text body not found]" if ($debug == 2); |
329 |
|
|
print STDERR "plain/text body not found\n" if ($debug); |
330 |
|
|
return; |
331 |
dpavlin |
25 |
} |
332 |
|
|
|
333 |
dpavlin |
14 |
# reformat with Text::Autoformat |
334 |
|
|
my $wrap = $self->{wrap_margin}; |
335 |
|
|
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
336 |
dpavlin |
21 |
$body = autoformat($body, {right=>$wrap, all=>1}); |
337 |
dpavlin |
14 |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
338 |
|
|
} |
339 |
|
|
|
340 |
|
|
return $body; |
341 |
dpavlin |
7 |
} |
342 |
|
|
|
343 |
|
|
|
344 |
dpavlin |
4 |
sub fetch_result_by_id { |
345 |
|
|
my $self = shift; |
346 |
|
|
|
347 |
|
|
my $id = shift || return; |
348 |
|
|
|
349 |
dpavlin |
13 |
my $row = $self->{cache}->{$id}; |
350 |
dpavlin |
2 |
|
351 |
dpavlin |
13 |
if (! $row) { |
352 |
dpavlin |
2 |
|
353 |
dpavlin |
13 |
print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2); |
354 |
dpavlin |
2 |
|
355 |
dpavlin |
17 |
my $message = $self->fetch_message($id) || return; |
356 |
dpavlin |
13 |
|
357 |
|
|
$row->{'id'} = $id; |
358 |
dpavlin |
19 |
|
359 |
dpavlin |
42 |
foreach my $p (@counters) { |
360 |
dpavlin |
22 |
foreach my $v ($self->unroll($message,$p,'phrase')) { |
361 |
dpavlin |
19 |
push @{$row->{$p}},$v; |
362 |
|
|
$self->add_counter($p,$v); |
363 |
|
|
} |
364 |
|
|
} |
365 |
dpavlin |
42 |
|
366 |
|
|
$self->add_counter('folder', $1) if ($id =~ m/^(\S+)\s/); |
367 |
|
|
|
368 |
dpavlin |
16 |
$row->{'subject'} = $self->decode_qp($message->subject); |
369 |
dpavlin |
13 |
$row->{'body'} = $self->plain_text_body($message); |
370 |
dpavlin |
20 |
my $utime = str2time($message->date); |
371 |
dpavlin |
13 |
|
372 |
dpavlin |
20 |
$row->{'date_utime'} = $utime; |
373 |
|
|
|
374 |
|
|
$row->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)); |
375 |
|
|
$self->add_counter_calendar($utime); |
376 |
|
|
|
377 |
dpavlin |
13 |
# XXX store in cache? |
378 |
|
|
$self->{cache}->{$id} = $row; |
379 |
|
|
print STDERR "$id stored in cache\n" if ($debug == 2); |
380 |
|
|
} else { |
381 |
|
|
print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2); |
382 |
dpavlin |
42 |
foreach my $p (@counters) { |
383 |
dpavlin |
23 |
foreach my $v (@{$row->{$p}}) { |
384 |
|
|
$self->add_counter($p,$v); |
385 |
|
|
} |
386 |
|
|
} |
387 |
|
|
|
388 |
dpavlin |
42 |
$self->add_counter('folder', $1) if ($id =~ m/^(\S+)\s/); |
389 |
|
|
|
390 |
dpavlin |
23 |
$self->add_counter_calendar($row->{date_utime}); |
391 |
dpavlin |
13 |
} |
392 |
|
|
|
393 |
dpavlin |
2 |
return $row; |
394 |
|
|
|
395 |
|
|
} |